1) The Transfer Modality Research Initiative (TMRI) in Bangladesh evaluated the impacts of different social protection interventions on household welfare and child nutrition. 2) The study found that cash transfers, food transfers, and combinations of the two all improved household consumption and food security. However, adding nutrition behavior change communication (BCC) led to significantly larger improvements. 3) In northern Bangladesh, cash transfers combined with BCC reduced child stunting rates by 7.3 percentage points over two years, suggesting social protection needs to address both income and nutrition knowledge to impact child nutrition outcomes.

1. LINKING SOCIAL PROTECTION AND
NUTRITION IN BANGLADESH
Results from the
Transfer Modality Research Initiative (TMRI)
Akhter Ahmed*, John Hoddinott**, Shalini Roy*, and Esha Sraboni***
*International Food Policy Research Institute, **Cornell University (former IFPRI researcher), ***Brown University (former
IFPRI researcher)
Presented by: Akhter Ahmed
Transform Nutrition: Evidence for Action in South Asia
Kathmandu, Nepal | 8 July 2017

2. Motivation
 While the decline in chronic undernutrition in Bangladesh is
impressive, the prevalence of stunting (36% in 2014) is still
very high.
 A key question: To improve child nutrition, are large-scale
social protection interventions that increase household
income sufficient?
 Most existing evidence (e.g., Ahmed et al. 2009 IFPRI study
in Bangladesh, IFPRI and other evaluations of CCTs in Latin
America) shows
Reduced household poverty and improved food security
But few improvements, if any, in child nutritional status
 Are there constraints other than income that also need to be
addressed, such as nutrition knowledge?
This is one of the focal set of issues addressed by the
Transfer Modality Research Initiative (TMRI) in
Bangladesh.

3. Objectives of the Transfer Modality
Research Initiative (TMRI)
 The overall objective of the research is to generate definitive
evidence on which safety net transfer modalities work best
for the ultra poor in rural Bangladesh.
 The research has the following specific objectives:
1. Measure the impact and cost-effectiveness of transfer
methods on these key outcomes:
– household income
– household food security
– child nutrition
2. Evaluate the process of delivering benefits at the
operational level and solicit feedback from program
participants.

4. TMRI – Basic Design
 Two-year (May 2012 – April 2014) project designed by
IFPRI and implemented by the UN World Food Programme
(WFP). Used randomized control trial with 5 treatment arms
and controls:
• Monthly cash transfer of Taka (Tk) 1,500 (about US$19
per month) (north & south)
• Monthly food transfer of 30 kg of rice, 2 kg of lentil pulse
and 2 liters of micronutrient fortified cooking oil (north &
south)
• ½ Food and ½ cash: Tk 750 and 15 kg of rice, 1 kg of
mosur (lentil) pulse and 1 liter of micronutrient fortified
cooking oil (north & south)
• Monthly cash transfer AND Nutrition Behavior Change
Communication (north)
• Monthly food transfer AND Nutrition Behavior Change
Communication (south)
• Controls (north & south)

5. Evaluating Impacts
 IFPRI developed a cluster randomized controlled trial (RCT)
design to evaluate impacts of transfer modalities.
 Randomly assigned 50 clusters (villages) to each of the five
groups (four treatment arms, one control) in each of the 2
regions (northwest and south).
 Selected 10 households from each village who met the
following criteria:(1) be poor; (2) have at least one child aged
0-24 months; and (3) not receive benefits from other safety
net interventions.
 Selected 500 clusters and 5,000 households (4,000
participants and 1,000 control households)
 We used RCT with “before-and-after” and “with-and-without”
differences for estimating the impact of transfers.
 We used the analysis of covariance (ANCOVA) regression to
estimate impact.

6. Household Surveys for Impact
Evaluation
 The required quantitative data for impact evaluation come
from three household surveys
 The first household survey, carried out in April 2012 (just
before the start of transfers), provided the information
needed for the baseline study
 A first follow-up survey was conducted in June 2013, just
after completing 12 months of transfer distributions
 A second follow-up or endline survey was conducted in April
2014, during the 24th month of transfer distribution
Panel data with low attrition (<3%) at endline, random
(not correlated with intervention assignment)

7. Adding BCC to transfers causes a
greater increase in “diet quantity” in
terms of household caloric intake
Absolute change (kcal/day)
75
143
83
282
0
50
100
150
200
250
300
Cash only Food only Cash+Food Cash+BCC
Increaseincalorie(kcal/person/day)
North
Statistically significant Not significant
141
23
8
38
0
50
100
150
200
250
300
Cash Food Cash+Food Food+BCC
Increaseincalorie(kcal/person/day)
South
Statistically significant Not significant

8. Adding BCC to transfers causes a greater
increase in per-calorie expenditure
Absolute change (taka)
1.80
0.94
1.20
4.35
0
1
2
3
4
5
Cash only Food only Cash+Food Cash+BCC
Per-caloriefoodexpenditure(Tk/1,000kcal)
North
Statistically significant Not significant
1.68
1.21
1.75
4.22
0
1
2
3
4
5
Cash Food Cash+Food Food+BCC
Per-caloriefoodexpenditure(Tk/1,000kcal)
South
Statistically significant Not significant

9. Calculation of the WFP Food
Consumption Score
(# of days consumed of each food group in past 7 days, weighted by
“nutritional importance”)

10. All modalities significantly increased
household diet quality in both regions:
Adding BCC gives a greater impact
(using WFP’s “Food Consumption Score”: 0-112)
6.9
9.1
6.9
23.7
0
5
10
15
20
25
Cash only Food only Cash+Food Cash+BCC
Foodconsumptionscore
North (baseline: 43.7)
Statistically significant Not significant
2.7
4.9 5.3
12.7
0
5
10
15
20
25
Cash only Food only Cash+Food Food+BCC
Foodconsumptionscore
South (baseline: 50.9)
Statistically significant Not significant

11. Adding BCC increases the frequency of
several food groups consumed by children
<42 months: North (impacts significant at ≤10% level)
6.1
7.3
11.7
6.4
24.6
10.9
22.8
36.0
15.1
0
5
10
15
20
25
30
35
40
Eggs Legumes Legumes Eggs Legumes Dairy
products
Flesh
foods
Eggs Vit A fruit
and veg
Cash Food Cash & Food Cash & BCC
Percentofchildrenwhoconsumedinpast24hours

12. Adding BCC increases the frequency of
several food groups consumed by children
<42 months: South
(impacts significant at ≤10% level)
7.3
10.8
8.0
25.1
12.7
15.1
20.0
0
5
10
15
20
25
30
35
40
Eggs Legumes Legumes Legumes Flesh foods Eggs Vit A fruit
and veg
Cash Food Cash & Food Food & BCC
Percentofchildrenwhoconsumedinpast24hours

13. Does child stunting reduce?
46.0
38.7
0
10
20
30
40
50
Baseline Cash + BCC
Childstuntingrate(percent)
North  In the north, “Cash +
BCC” improved child
nutritional status in terms
of reducing the
prevalence of stunting by
7.3 percentage points,
but other modalities did
not.
 No treatment arm had
any impact on any
measure of
anthropometric status of
children in the south.

14. Conclusions
 The overall story that emerges is that all transfer modalities
in both northwest and southern regions cause meaningful
improvements in nearly all measures of consumption.
 However, the addition of nutrition behavior change
communication to transfers consistently causes much larger
improvements than transfers alone.
 In northwestern Bangladesh, cash transfers combined with
nutrition BCC led to a decrease of 7.3 percentage points in
child stunting over the two years of the project – an
achievement almost three times the national average
decline.

15. Policy Implications
 If policy objective is to improve the diets of poor households,
both cash and food transfers are effective.
 If policy objective is to improve the nutritional status of
children from the poorest households, transfers alone are
inadequate.
 High quality Behavior Change Communication together with
transfers – especially cash transfers – appear to deliver
large improvements into child nutrition and anthropometric
outcomes.
 Although the impacts of cash and food transfers on most
outcomes are quite similar, the delivery cost of cash transfer
is considerably lower than that of food transfer. Therefore,
cash transfer appears to be more cost-effective than food
transfer.

16. Policy Uptake
 Evidence from the TMRI is prompting the Bangladesh
government and development partners to consider adding a
nutrition BCC to social protection programs.
 For example, encouraged by TMRI results, the Ministry of
Women and Children Affairs (MoWCA) is piloting the
Investment Component for Vulnerable Group Development
(ICVGD) program for 8,000 women, which adds a cash
grant for investment, fortified rice distribution, and nutrition
BCC to existing VGD activities.